View this file with results and syntax highlighting here.
Here's a little BQN program:
#! /usr/bin/env bqn
# Case conversion utilities
case ← {
diff ← -´ "Aa"
Lower ⇐ -⟜diff
Upper ⇐ Lower⁼
}
hw ← <˘ 2‿∘ ⥊ "helloworld"
hw case.Upper⌾(⊑¨)↩
•Out hw ↩ ∾ ⥊⍉ [hw, ", "‿"!"] # Hello, World!
# Split at spaces and repeated characters
Split ← {
!1==𝕩 ⋄ (!2=•Type)¨𝕩
Proc ← {
· 𝕊 ' ': spl⇐1 ; # Space: break and delete it
prev Fn cur: ⟨spl,str⟩⇐
spl←0 ⋄ str←⟨cur⟩ # Include and don't break...
{ prev=cur ? spl+↩1 ; @ } # except at equal characters
}
GV‿GS ← {𝕏¨}¨ ⟨ {⟨s⇐str⟩:s;""}
{𝕩.spl} ⟩
r ← Proc{»𝔽¨⊢} 𝕩
(∾¨ GV ⊔˜ ·+`GS) r
}
•Show Split hw # ⟨ "Hel" "lo," "World!" ⟩
It's not the most idiomatic BQN you'll see, but that's because this piece of code uses nearly all the syntax in the language (and a good number of the primitives).
If you save it with the name hello.bqn and have BQN installed, the script can be run with $ bqn hello.bqn from a shell. Because of the #! line at the top, $ ./hello.bqn also works if bqn is in your path and hello.bqn is executable. It can also be run from another BQN file in the same directory, or REPL started there, using •Import "hello.bqn". Or just copy-paste it into the online REPL.
Now let's see how it works.
# Case conversion utilities
case ← {
diff ← -´ "Aa"
Lower ⇐ -⟜diff
Upper ⇐ Lower⁼
}
This part of the code defines a namespace using braces {}, then assigns it to the name case. There are three assignments inside the namespace too. Since BQN uses lexical scoping, code outside the namespace can't access the variables diff, Lower, and Upper directly. Oh, and the first line is a comment.
The value diff is the result of applying a function -´ to the argument "Aa". Function application is always written just by placing a function next to its arguments like this—a prefix application if there's one argument, infix if there are two, and that's the most arguments you can have. This doesn't limit BQN's capabilities because it's easy to pass a list as an argument. In fact, "Aa" is a string, which means a list of characters. Characters are written with single quotes, so it's a list of 'A' and 'a'.
-´ "Aa" # Prefix application
'A' - 'a' # Infix application
The function -´ is a compound function, because it consists of another function - (it's just subtraction) passed to a 1-modifier Fold (´). Fold applies its operand function - between the elements of its list argument "Aa". For a more familiar example, +´ could sum a list of numbers. But here we end up taking the difference between two characters, an instance of character arithmetic. Characters are always Unicode code points, and this operation takes the difference between their numeric values. The important point is that the difference between any lowercase Latin character and its uppercase version is always the same number (specifically -32, which is written ¯32 because ¯ can be part of a numeric literal while the function - can't). Adding this number to a lowercase letter translates it to uppercase:
diff ← -´ "Aa"
'b' + diff
The function Lower is defined to be -⟜diff, but it uses a different arrow ⇐ to do this. This is an export, and it declares that Lower is a field of a namespace that can be accessed from the outside. Having a ⇐ in it is also what makes the block define a namespace. Lower itself won't be used for a while, but Upper is accessed a few lines down, with case.Upper.
Lower is created with a modifier again, this time the 2-modifier ⟜. We've now seen one each of the three primitive types: function, 1-modifier, and 2-modifier. There are a lot of primitives, but some simple rules tell you which type they have. Primitives are functions by default, but superscript characters are 1-modifiers (´⁼˘¨˜` in our program), and ones with an unbroken circle are 2-modifiers (⟜∘⌾; ⍉ is a broken circle so it's a function). Variable names follow a similar system, where functions start with an uppercase letter and subjects with a lowercase one.
After (⟜) takes two operands, - and diff, to produce a function—specifically, it binds diff as the right argument of -, so that calling it on an argument subtracts diff from that argument.
-⟜diff 'G'
'G' - diff
Upper could be written the same way, with +⟜diff or diff⊸+. Instead, it's defined to be the inverse of Lower with Undo (⁼). BQN knows a few algebraic tricks to invert primitives and compound functions, and -⟜diff is well within its capabilities.
The next part of the program begins to use BQN's array-oriented capabilities. It consists of three statements, which BQN evaluates in order (after the previous statement, which defined case).
hw ← <˘ 2‿∘ ⥊ "helloworld"
hw case.Upper⌾(⊑¨)↩
•Out hw ↩ ∾ ⥊⍉ [hw, ", "‿"!"] # Hello, World!
The first one takes the string "helloworld" and turns it into a list of two strings.
<˘ 2‿∘ ⥊ "helloworld"
This is our first expression that evaluates two functions, and has a lot of other stuff going on besides. Let's draw out the expression grouping. The online REPL can also do this for you if you click "Explain" before running your expression.
The two functions evaluated are <˘ and ⥊. Functions all have the same precedence, and are evaluated from right to left—the same order as f(g(h(x))) in math but without the parentheses. Applying a modifier like ˘ has higher precedence than function application, and so does stranding ‿. Stranding is a quick way to write a list; 2‿∘ could also be written ⟨2,∘⟩ using the brackets found later in the program. Elements can be anything: here 2 is a list and ∘ is a 2-modifier.
2‿∘
The first function applied is Reshape, which reshapes its right argument according to the shape on the left. So a shape of 2‿5 would turn our 10-character list into a 2×5 array. But 2‿∘ isn't really a shape: the ∘ indicates a computed length to be filled in based on the size of the array being reshaped.
2‿5 ⥊ "helloworld"
2‿∘ ⥊ "helloworld"
This result is a 2-dimensional array of characters. A list is also a kind of array, but with only one dimension. We say a list has rank 1, while the new array has rank 2. But the next thing we do is turn our array into a list of lists:
<˘ 2‿∘ ⥊ "helloworld"
The function that does this is Enclose Cells, <˘. The Cells modifier means we're working with major cells, which are the parts of an array with one less dimension. For a rank-2 array, these are its rows, while for a list, they're rank-0 arrays (or units), each containing a single element. Enclose Cells applies Enclose to each major cell of its argument to produce the major cells of its result. Enclosing a cell wraps it up in a rank-0 array, and using these for the result's major cells creates a list, whose elements are the rows of the rank-2 array. What happens to this list of lists next?
hw case.Upper⌾(⊑¨)↩
This statement consists of the name hw just defined, a compound function, and then the new character ↩. This is another form of assignment, like ←, but it changes the value of an existing variable instead of defining a new one. There's also some special ↩ syntax here: the expression val Fn↩ is shorthand for val ↩ Fn val, avoiding the need to write the name hw twice (and val Fn↩ arg means val ↩ val Fn arg, like += and so on from C). So we are modifying hw by applying this function case.Upper⌾(⊑¨).
hw ← <˘ 2‿∘ ⥊ "helloworld"
case.Upper⌾(⊑¨) hw
hw case.Upper⌾(⊑¨)↩ # Sets new value for hw
That converts the first character of each string to uppercase! case.Upper is the case conversion function defined before, so that part makes sense. The rest of the function, ⌾(⊑¨), would be pronounced "Under the First of Each", which… pretty much makes sense too? The First Each function extracts the first element of each list in hw, the part that used to be "hw" but is now "HW".
⊑¨ hw
case.Upper "hw"
The Under modifier keeps track of where that string came from and puts it back, to produce a new, altered array. It's kind of special, like Undo, but works on all sorts of fancy selections. It's also worth pointing out that case.Upper applies to a string here, not an individual character. That's because arithmetic is pervasive, so that functions made of arithmetic naturally work on arrays. Although in this case it wasn't really necessary, because it's also possible to map over the two strings and uppercase the first character of each separately:
case.Upper⌾⊑¨ "hello"‿"world"
Modifiers are applied from left to right, opposite to functions (1-modifiers also take the operand on the left while prefix functions have the argument on the right). So case.Upper⌾⊑¨ means (case.Upper⌾⊑)¨.
The variable hw is modified one more time, then printed, producing the output Hello, World!
•Out hw ↩ ∾ ⥊⍉ [hw, ", "‿"!"] # Hello, World!
None of these functions have a subject to the left, so they're all evaluated as prefix functions. But first, we have the array notation []. This syntax forms an array from its major cells hw and ", "‿"!" (another strand, a list of two strings). Because the major cells are both lists, we have another rank 2 array.
[hw, ", "‿"!"]
The reason for forming this array is to interleave the elements, or we might say to read down the columns rather than across the rows. This ordering is done with a Transpose to exchange the two axes, then a Deshape to flatten it out, giving a list.
⍉ [hw, ", "‿"!"]
⥊ ⍉ [hw, ", "‿"!"]
Finally, Join combines this list of strings into a single string.
hw ↩ ∾ ⥊⍉ [hw, ", "‿"!"]
hw
The full statement stores this back in hw with ↩, then prints it using •Out. Assignment can be used inline, much like a function! •Out is our first system function (see this section), and prints a string directly as output. We have now printed that which new programmers must print, and covered the basics of BQN expressions!
Now we're going to play around with the string hw or "Hello, World!" that we've constructed, and see a few ways to construct functions. If you're starting out you won't need many of the details here for a while, so you may want to stop after getting the basic idea and revisit this page later.
# Split at spaces and repeated characters
Split ← {
!1==𝕩 ⋄ (!2=•Type)¨𝕩
Proc ← {
· 𝕊 ' ': spl⇐1 ; # Space: break and delete it
prev Fn cur: ⟨spl,str⟩⇐
spl←0 ⋄ str←⟨cur⟩ # Include and don't break...
{ prev=cur ? spl+↩1 ; @ } # except at equal characters
}
GV‿GS ← {𝕏¨}¨ ⟨ {⟨s⇐str⟩:s;""}
{𝕩.spl} ⟩
r ← Proc{»𝔽¨⊢} 𝕩
(∾¨ GV ⊔˜ ·+`GS) r
}
•Show Split hw # ⟨ "Hel" "lo," "World!" ⟩
The big definition Split is a block function, using {} like the namespace case—that was an immediate block. The difference is that Split contains an 𝕩, which indicates an argument. We also see that blocks can be nested within each other. The inner blocks contain other characters like ⇐ and 𝔽 that can change the nature of a block, but these only affect the block immediately containing them.
To begin with, Split tests its argument 𝕩. There are two tests, with a statement separator ⋄ between them. The diamond, as well as ,, is interchangeable with a newline. The tests are done with the function Assert (!).
!1==𝕩 ⋄ (!2=•Type)¨𝕩
First, Split requires that the rank =𝕩 be 1, that is, 𝕩 must be a list. In 1==𝕩, = has two meanings, depending on whether it has a left argument. Next, it checks that each element has a character type.
The subexpression !2=•Type is a function train, and it happens to have a simple expansion, as (!2=•Type)e is !2=•Type e. It matters that 2 is just a number; if it were a function it would be applied to e. We'll discuss trains more later. This one is applied with Each to test every element of 𝕩. This does form an array result, but it's not used.
The function Proc is called on each character of 𝕩 along with the previous character, and says what to do at that position. Also, it's deliberately inconsistent to cover more BQN features. Here's the whole thing:
Proc ← {
· 𝕊 ' ': spl⇐1 ; # Space: break and delete it
prev Fn cur: ⟨spl,str⟩⇐
spl←0 ⋄ str←⟨cur⟩ # Include and don't break...
{ prev=cur ? spl+↩1 ; @ } # except at equal characters
}
Unlike Split, which only contains a sequence of statements, Proc has some structure. Let's reduce each block body to a … to see it better.
Proc ← {
· 𝕊 ' ': … ;
prev Fn cur: …
}
This function has two bodies with ; in between. Each one has a header, separated from the body with a :. A header indicates the kind of the block as a whole, and also which inputs the body after it works on. It mirrors the way the block should be used. In the right context, both · 𝕊 ' ' and prev Fn cur would be valid function calls. Which tells us Proc is a function.
The first header, · 𝕊 ' ':, is more specific. The function is unlabelled, since 𝕊 just indicates the block function it's in (useful for recursion). The right argument is ' ', a space character, so this body will only be used if 𝕩 is a space. And the left argument is… ·, which is called Nothing. Both here and as an assignment target, Nothing indicates an ignored value. This body does require a left argument, but it doesn't name it. And the body itself is just spl⇐1. The ⇐ makes this body (only this one!) return a namespace, which has only the field spl.
The next header, prev Fn cur:, sets names for the function and its arguments, but doesn't constrain them other than requiring two arguments. So it applies in all the cases where the previous one didn't match, that is, when 𝕩 isn't ' '. The body starts with ⟨spl,str⟩⇐, and the ⇐ means it will return a namespace too. This is an export statement, which declares spl and str to be fields but doesn't define them—they must be defined somewhere else in the block, which is what happens next.
prev Fn cur: ⟨spl,str⟩⇐
spl←0 ⋄ str←⟨cur⟩ # Include and don't break...
{ prev=cur ? spl+↩1 ; @ } # except at equal characters
Both ⟨spl,str⟩ and ⟨cur⟩ are written with list notation, giving a list of two names, then one value. While ⟨spl,str⟩ can also be written spl‿str, there's no way to write ⟨cur⟩ with stranding.
On the last line we're now three blocks deep! This block also has two bodies, but they don't have headers. A predicate prev=cur ? tests whether to use the first body, which increments spl with modified assignment. Note that 𝕨=𝕩 wouldn't work here, because the special names 𝕨 and 𝕩 pertain only to the surrounding block, and Proc is a level up. However, the idiomatic way to write this part would be the much shorter spl←prev=cur, since BQN's booleans are 0 and 1.
The end result of Proc is always a namespace. It has a field spl set to 1 if 𝕩 is ' ' or the two arguments are equal. And if 𝕩 isn't ' ', it has another field str set to ⟨𝕩⟩.
Once Proc is applied to all the characters, we'll end up with a list of namespaces (which, yes, is over-engineered for what we're doing). The following statement defines two functions GV and GS to extract fields from this list.
GV‿GS ← {𝕏¨}¨ ⟨ {⟨s⇐str⟩:s;""}
{𝕩.spl} ⟩
Going left to right, GV‿GS indicates destructuring assignment, which will expect a list of two values on the right and take it apart to assign the two names. The right hand side is the function {𝕏¨}¨ applied to a list.
{𝕏¨} is a block function, like Split but a lot shorter. It uses the uppercase 𝕏 instead of 𝕩, so that it treats 𝕩 as a function (it doesn't require it to be a function, though: see mixing roles). It adds an Each ¨ onto its argument. This is used to convert the two functions in the list from functions that work on a namespaces to functions that work on a list of them.
The list is split across two lines, using newline as a separator instead of , or ⋄. Its second function {𝕩.spl} is simpler: it takes a namespace 𝕩 and gets the field named spl.
The first function is more complicated, because the argument namespace might or might not have an str field. The list-like notation ⟨s⇐str⟩ is another example of destructuring assignment, but this time it destructures a namespace, using an alias to give it a short name. This header leaves off the function name 𝕊, using a special rule for one-argument functions. Arguments in headers are very similar to assignment targets, but if the destructuring doesn't match it tries the next body (if there is one) instead of giving an error. So if the argument is a namespace with an str field then {⟨s⇐str⟩:s;""} returns that field's value, and otherwise it returns "".
Now that Split has defined Proc, GV (get value), and GS (get split), it's ready to do its work.
r ← Proc{»𝔽¨⊢} 𝕩
(∾¨ GV ⊔˜ ·+`GS) r
The first line here applies Proc to each character and the one before it, using ' ' as the character "before" the first. Proc{»𝔽¨⊢} 𝕩 is a fancy way to write (»𝕩) Proc¨ 𝕩, which we'll explain in a moment. First, here's what the Nudge function » does.
hw
»hw
It moves its argument forward by one, adding a space character (the array's fill) but keeping the same length. This gives the previous characters that we want to use for Proc's left argument. Here Each is used with two arguments, so that it iterates over them simultaneously, like a "zip" in some languages.
What about the fancy syntax Proc{»𝔽¨⊢} 𝕩? The block {»𝔽¨⊢} is an immediate 1-modifier because it uses 𝔽 for an operand but not the arguments 𝕨 or 𝕩. This means it acts on Proc only, giving »Proc¨⊢, which is a train because it ends in a function ⊢. Following the rules for a 3-train, (»Proc¨⊢)𝕩 expands to (»𝕩) Proc¨ (⊢𝕩), and since ⊢ is the identity function, ⊢𝕩 is 𝕩.
Since a display of lots of namespaces isn't too enlightening, we'll skip ahead and show what the results of GV and GS on those lists would be. GV turns each character into a string, except it makes a space into the empty string. GS has a 1 in the places where we want to split the string.
sp ← ' '=hw
gv ← (1-sp) ⥊¨ hw
gs ← sp ∨ »⊸= hw
gv
gs
(∾¨ GV ⊔˜ ·+`GS) r
The next part is a bigger train. Trains are grouped into threes starting at the end, which takes some time to get used to. Here's a diagram showing how this one works.
There are actually three train groupings here: from right to left, ·+`GS, GV ⊔˜ …, and ∾¨ …. Two of them are 2-trains, which apply one function to the result of another, but the one with ⊔ is a 3-train, applying a function to two results. In the end, functions GS and GV are applied to r. In fact, to evaluate the entire train we can replace these two functions with their results, giving ∾¨ (GV r) ⊔˜ ·+`(GS r).
∾¨ gv ⊔˜ ·+`gs
In this expression, Nothing can be removed without changing the meaning. It's used in the train to force +` to apply to GS as a 2-train instead of also taking ⊔˜ as a left argument. The Scan +` is a prefix sum, progressively adding up the numbers in gs.
gs
+`gs
The next bit uses Swap to switch the order: gv ⊔˜ +`gs is (+`gs) ⊔ gv, but sometimes the different order can read better (here it was mostly to squeeze Nothing into the program, I'll admit). Group then splits gv up based on the indices given: the first three elements become element 0 of the result, the next three element 1, and the rest element 2.
(+`gs) ⊔ gv
Then Join Each uses two functions we've seen before to build the final result!